The Digital Farms of British Columbia
A screening model, not a bankable feasibility study.
The website uses transparent first-pass energy and employment assumptions so readers can see what would have to be true for the concept to work.
Core screening assumptions
| Variable | Screening value | Why it matters |
|---|---|---|
| Annual operating hours | 8,760 h | Converts MW into annual energy. |
| Representative IT load factor | 85–95% | AI loads are high, but not perfectly flat. |
| Recoverable useful heat | 65–80 GWh/y per 10 MW IT block | Allows for capture and distribution losses. |
| Permanent direct roles | 35–55/site Phase I; 45–75/site Phase II | Combines compute and agricultural operations. |
| Primary power | BC grid | Solar is supplemental, not the firm resource. |
| Water cooling | No once-through river cooling assumed | Reduces hydrological and thermal-discharge risk. |
What can break the model?
Insufficient grid capacity, expensive substation upgrades, lack of carrier diversity, seasonal mismatch between server heat and customers, land-use restrictions, agricultural water constraints, flood or wildfire risk, community opposition, or a failure to secure commercially viable compute customers.
What makes it stronger?
Sites with existing greenhouses or processors, short heat-pipe distances, staged utility connections, multiple fibre carriers, strong local workforce capacity, and counterparties willing to sign long-term heat or food agreements.
Next engineering work
- Utility pre-application and feeder/substation screening for every node.
- Carrier and dark-fibre route survey.
- Heat-demand profiles by hour and month for candidate users.
- Parcel screening: ALR, flood, wildfire, geotechnical, road, noise and setbacks.
- First Nations and municipal engagement before preferred-site selection.
- Class 4/3 capital estimate and levelized heat cost.